Shrike

Provides a secure scan pipeline for prompts, responses, SQL, files, and web searches to prevent prompt injection and data leakage.

typescript

GitHub Stars

typescript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "shrike-security-shrike-mcp": {
      "command": "npx",
      "args": [
        "shrike-mcp"
      ],
      "env": {
        "MCP_DEBUG": "false",
        "SHRIKE_API_KEY": "YOUR_API_KEY_OR_EMPTY_IF_USING_FREE_TIER",
        "SHRIKE_BACKEND_URL": "https://api.shrikesecurity.com/agent",
        "MCP_SCAN_TIMEOUT_MS": "15000",
        "MCP_RATE_LIMIT_PER_MINUTE": "100"
      }
    }
  }
}

You deploy Shrike MCP Server to protect AI agents from prompt injections, jailbreaks, and other risky operations. You can run it locally or integrate it with compatible clients to scan prompts, responses, SQL, files, and web searches, leveraging a multi-layer security pipeline that you can enable with an API key for enhanced analysis.

How to use

You use an MCP client to connect to Shrike MCP Server, run scans, and receive structured results that indicate whether content is allowed or blocked. You can start with a local or hosted MCP setup and configure clients to pass prompts, responses, and related data for analysis. Typical workflows include scanning user prompts before processing, validating LLM outputs, and inspecting SQL queries or file writes for dangerous patterns. When you provide an API key, you gain access to advanced layers that include semantic analysis and expanded threat detection.

How to install

Prerequisites: you should have Node.js and npm installed on your system.

Install Shrike MCP Server globally so you can run it from any directory.

npm install -g shrike-mcp

Or run Shrike MCP Server directly without installation using npx.

npx shrike-mcp

Additional sections

Configuration and startup patterns let you tailor Shrike MCP Server to your environment. You can run the server in standard input/output mode and point clients to the running MCP process. The server supports environment variables to customize behavior and connect to the Shrike backend for enhanced analysis.

Available tools

scan_prompt

Scans user prompts for prompt injection, jailbreak attempts, and malicious content. Supports PII redaction with token-based rehydration.

scan_response

Scans LLM-generated responses for leaks, PII exposure, toxicity, and topic drift, with potential rehydration of PII tokens.

scan_sql_query

Scans SQL queries for injection attacks and dangerous operations before execution.

scan_file_write

Validates file write operations, checking for path traversal, secrets, and sensitive file access.

scan_web_search

Scans web search queries for PII exposure, exfiltration risks, and blocked domains.

report_bypass

Reports content that bypassed security checks to improve detection and threat learning.

get_threat_intel

Retrieves current threat intelligence, including active patterns and statistics.

Built by

VeilStrat

AI signals for GTM teams